Although highly active antiretroviral therapy (HAART) has a dramatic initial effect in treating HIV-disease, eventually it has a failure rate of up to 60%. Two potential sources of HAART failure will be examined: patients beginning HAART with pre-existing resistance-conferring mutations and patients? nonadherence to the complex drug regimen. HAART is a chronic therapy with multiple potentially toxic drugs each often having their own individual dosing schedules and requirements. Add to this a disadvantaged patient population with a lack of ready access to and ability to afford an expensive regimen and adherence becomes problematic. We hypothesize that nonadherence will be the primary cause of failure of HAART among treatment-naive HIV-infected patients. Other patients have received mono-or dual nucleoside therapy for HIV, allowing the virus to build up a sequential resistance. We further hypothesize that prior exposure to drug therapy with resultant resistance-conferring mutations will be the primary cause of HAART failure in patients already having received treatment. Upon entry into this protocol, an initial blood sample will be taken for lymphocyte profile, viral load, and characterization of genotypic resistance. To determine the initial resistance profile, the reverse transcriptase and protease genes of plasma HIV will be sequenced and checked for the presence of mutations at 53 sites that have been associated with primary or secondary drug resistance. At six month intervals samples from follow-up visits will be tested for incomplete suppression of plasma viral RNA load (<50 copies/ml). Patients with incomplete suppression will be scheduled for a confirmatory sample. This confirmatory sample will be tested again for viral load, sequenced for resistance mutations, and as an objective measure of adherence, analyzed for plasma drug concentrations. At each six month visit, the patient will be given a questionnaire inquiring about the ability to adhere with the therapeutic regimen and allowing the patient to specify and rank the reasons why adherence is difficult. The patient will be asked to make suggestions on how adherence could be made easier. At the end of the study (3 years) all patients will be tested for lymphocyte profile, viral loads, resistance-conferring mutations, and drug concentration. The inability of HAART to suppress viral appearance will be correlated to the presence of resistance-conferring mutations at the beginning of the study, to initial viral load, and to markers of the adherence. These results and feedback from the patients will help elucidate the causes of HAART failure and help improve strategies for ensuring regimen adherence.